Research
Nine domains of Artificial Space Intelligence research
Our programme is organised into nine interconnected research domains. Each describes questions we are investigating and capabilities we believe are achievable — not products in service today.
Models and architectures designed for space conditions — radiation tolerance, tight power budgets, sparse data and no room for silent failure.
Planning, fault detection and recovery that let spacecraft act correctly when the ground link is unavailable.
Turning raw downlink into decision-ready insight, and moving parts of that processing onto the satellite itself.
Compute in orbit: scheduling, thermal and power constraints, distributed inference across a constellation.
Perception and manipulation for servicing, assembly and surface operations where teleoperation is impractical.
Multi-sensor fusion and change detection to measure what is happening on the surface, at scale and over time.
Tracking, conjunction assessment and debris risk modelling for an increasingly congested orbital environment.
Autonomy for cislunar and planetary missions: navigation, resource mapping and long-delay decision making.
Governance, safety and ethics of autonomous systems that observe Earth and act on behalf of people.
Method
How a research direction becomes a capability
Each domain moves through the same sequence: define the question, model the environment, simulate, validate against real mission data where available, then publish what we learn.
Stage 1
Question
Stage 2
Model & simulate
Stage 3
Validate
Stage 4
Publish